Cisco Targets AI Trust with Galileo Deal

If the original Galileo spent his time figuring out how things move and fall, Cisco is now tackling a version of that problem in AI, trying to understand how these systems behave once set loose.

The company announced plans to acquire Galileo Technologies, an AI observability startup focused on helping enterprises monitor and evaluate how their AI systems behave in real time.

The focus is on what these systems are actually doing once they’re live.

Why Galileo focuses on the trust layer in AI deployment

That’s where Galileo comes in. The platform focuses on tracking AI systems as they’re built and used, with an emphasis on catching issues early and putting guardrails in place for how those systems operate. 

They’re not worried about the traditional observability signals; this is more about understanding the output itself.

“Galileo was purpose-built to solve one of the hardest and most consequential problems in AI: Trust,” wrote Kamal Hathi, senior vice president and general manager of Cisco’s Splunk business. “From day 1, its platform has given AI teams the tools to evaluate AI quality, detect AI failures before they reach users, and continuously improve AI behavior in production – turning observability from a nice-to-have into a core pillar of AI development.”

Extending observability into the AI stack

The plan is to bring Galileo into Cisco’s Splunk observability platform, with a focus on monitoring AI agents. It tracks what happens across the lifecycle, from early model decisions through to what’s actually running in production.

That lifecycle is already pretty layered, especially as teams work with multiple agents that can operate with some independence. So the focus isn’t just on whether something is running, it’s also on what those systems are actually producing, including things like hallucinations, bias, security risks, and how usage and cost are tracking over time.

“The behavior of agentic applications can lead to unexpected, inaccurate, low quality, or harmful outputs,” Hathi wrote. “As a result, teams need visibility across the AI stack beyond signals like latency and errors.”

Cisco says Galileo will provide a single platform to instrument each stage of that lifecycle, adding deeper evaluation and control into workflows that are still evolving.

A broader push toward AI trust and control

The deal also lines up with Cisco’s recent push to position itself as a trust layer for AI systems. The company has introduced products to identify and monitor AI agents and expand security protections around them.

Galileo adds another piece to that strategy. Instead of focusing solely on network or identity layers, Cisco is moving deeper into how AI systems behave at the application and workload levels.

For teams building or managing AI systems, getting models into production isn’t really the hard part anymore. Keeping them predictable and accountable once they’re out there is where things tend to break.

The acquisition is expected to close in the fourth quarter of Cisco’s fiscal year 2026. Until then, both companies will operate independently.

Cynomi just rolled out a set of AI-powered CISO agents aimed at MSPs, essentially baking security decision-making and oversight directly into software. That lines up pretty neatly with what Cisco is doing here. As more of this work gets handed off to agents, the real question becomes how you keep tabs on what they’re doing and whether it’s actually the right call.